Integrating gut physicochemical gradients and mobile genetic element dynamics to decipher horizontal gene transfer in termite microbiomes

Abstract

Abstract Mobile genetic elements drive horizontal gene transfer (HGT), which may have contributed to the functional diversity of termite gut microbiomes, although the environmental determinants of these processes remain poorly understood. Our proposed model integrates spatially resolved physicochemical gradients and mobile genetic element (MGE) dynamics to quantitatively predict HGT patterns in termite guts. The framework integrates high-resolution metabolomic and transcriptomic profiling data from gut compartments, sourced from published experimental studies, with a generalized additive model to forecast the rate of MGE transfer as a function of local environmental conditions, and goes beyond static cataloguing of HGT events. Moreover, metabolic network alignments have been used to predict MGE-host interactions based on functional complementarity, and a heterogeneous graph neural network is used to dynamically predict HGT hotspots using environmental, MGE, and microbial host data. Unlike traditional methods, our framework explicitly links microenvironmental variability with gene flow mediated by MGEs using taxonomic, metagenomic, and metabolic features. The framework incorporates state-of-the-art computational methods, such as nanoscale secondary ion mass spectrometry for spatial metabolite mapping and graph attention networks (GAT) for transfer-likelihood prediction. The study of physicochemical gradients in the context of MGE-host interactions offers a mechanistic framework for investigating HGT dynamics in complex microbial communities, with broader implications for understanding gene flow in other host-associated ecosystems.